Tencent

Every Question Has Its Own Value: Reinforcement Learning with Explicit Human Values

ReLook: Vision-Grounded RL with a Multimodal LLM Critic for Agentic Web Coding